An Assessment of VAS-Derived Retrievals and Parameters Used in Thunderstorm Forecasting

Abstract

Operational VAS satellite retrievals and derived parameters used in forecasting severe local storms were evaluated against corresponding radiosonde values. VAS products also were compared with the first-guess input to the retrieval algorithm. The evaluation methodology was to pair each radiosonde observation (RAOB) with the closest VAS retrieval within 50 km during a 4-month period in 1986. VAS temperatures wore found to agree closely with radiosonde values; however, VAS dewpoints showed somewhat less agreement. VAS/RAOB sounding differences were poorly correlated with the number of pixels from which the retrievals were prepared. VAS discrepancies usually wore not well correlated with first-guess errors. A disappointing finding is that the retrievals degraded the first guess about as often as they improved it. VAS did tend to improve large first-guess errors, but not even this was guaranteed. Horizontal gradients of VAS products generally were stronger than those from radiosondes. VAS precipitable water agreed better with the ground truth than did dewpoints at individual levels, but VAS thicknesses were not much improved over the already accurate VAS temperatures. Results for the severe storms forecasting parameters indicated that VAS/RAOB discrepancies increased with the amount of manipulation required during computation. VAS parameters incorporating observed surface data tended to give better results than those that did not. Of all the parameters examined, VAS-derived lifted index exhibited the best agreement with RAOB versions. VAS positive buoyant energy showed disappointingly poor comparisons. VAS retrievals provided poor measures of the low-level negative buoyant energy that must be overcome before convection can begin. Agreements between VAS/RAOB versions of the K index, total totals index, Showalter index, and several other parameters were intermediate to those of positive buoyant energy and lifted index.